The study highlights the limitations of existing Human-Machine Interfaces (HMI) that use surface electromyography (sEMG) for gesture recognition, specifically in applications like prosthetics and rehabilitation.
A new strategy utilizing wearable A-mode ultrasound and a two-stage cascade model is introduced, which effectively classifies grasping gestures while simultaneously estimating applied force.
Experimental results show that this new method outperforms traditional models in both classification and force estimation, achieving fast real-time recognition suitable for practical applications.